cancer cell article - kevan shokatshokatlab.ucsf.edu/pdfs/18691552.pdfcancer cell article discovery...
TRANSCRIPT
Cancer Cell
Article
Discovery of Drug-Resistantand Drug-Sensitizing Mutationsin the Oncogenic PI3K Isoform p110a
Eli R. Zunder,1 Zachary A. Knight,2,6 Benjamin T. Houseman,3 Beth Apsel,2 and Kevan M. Shokat3,4,5,*1Graduate Group in Biophysics2Chemistry and Chemical Biology Graduate Program3Department of Cellular and Molecular Pharmacology4Howard Hughes Medical Institute
University of California, San Francisco, San Francisco, CA 94158, USA5Department of Chemistry, University of California, Berkeley, Berkeley, CA 94720, USA6Present address: Howard Hughes Medical Institute, The Rockefeller University, New York, NY 10021, USA*Correspondence: [email protected]
DOI 10.1016/j.ccr.2008.06.014
SUMMARY
p110a (PIK3CA) is the most frequently mutated kinase in human cancer, and numerous drugs targeting thiskinase are currently in preclinical development or early-stage clinical trials. Clinical resistance to protein ki-nase inhibitors frequently results from point mutations that block drug binding; similar mutations in p110a arelikely, but currently none have been reported. Using a S. cerevisiae screen against a structurally diverse panelof PI3K inhibitors, we have identified a potential hotspot for resistance mutations (I800), a drug-sensitizingmutation (L814C), and a surprising lack of resistance mutations at the ‘‘gatekeeper’’ residue. Our analysis fur-ther reveals that clinical resistance to these drugs may be attenuated by using multitargeted inhibitors thatsimultaneously inhibit additional PI3K pathway members.
INTRODUCTION
p110a is a class IA phosphatidylinositol 3-kinase (PI3K) that
phosphorylates phosphatidylinositol 4,5-bisphosphate (PIP2) at
the D-3 position of its inositol ring, generating phosphatidylinosi-
tol-3,4,5-trisphosphate (PIP3) (Fruman et al., 1998). PIP3 recruits
downstream signaling proteins such as Akt to the plasma mem-
brane, leading to increased growth, proliferation, motility, and
cell survival (Cantley, 2002).
p110a has attracted considerable interest as a drug target
(Hennessy et al., 2005; Luo et al., 2003; Stephens et al., 2005)
since its identification as an oncogene (Chang et al., 1997) and
the discovery of activating point mutations in its encoding
gene, PIK3CA (Samuels et al., 2004), as it is the most frequently
mutated oncogene in breast (27%) and endometrial (22%) can-
cers (frequencies obtained from the Wellcome Trust Sanger
180 Cancer Cell 14, 180–192, August 12, 2008 ª2008 Elsevier Inc.
Institute Cancer Genome Project; http://www.sanger.ac.uk/
genetics/CGP). In the absence of p110a mutation, other PI3K
pathway members, including receptor tyrosine kinases (RTKs),
Ras, PTEN, and Akt, are frequently mutated instead (Brugge
et al., 2007; Cully et al., 2006; Shaw and Cantley, 2006), high-
lighting the critical role that PI3K signaling plays in oncogenesis
and tumor maintenance (Gupta et al., 2007; Lim and Counter,
2005). More than 80% of reported p110a mutations are confined
to the helical domain mutations E542K and E545K and the kinase
domain mutation H1047R (Samuels et al., 2004). These mutants
exhibit increased catalytic activity (Ikenoue et al., 2005; Samuels
et al., 2004) and are capable of transforming a number of cell
lines in tissue culture and xenograft models (Ikenoue et al.,
2005; Isakoff et al., 2005; Kang et al., 2005; Samuels et al.,
2005; Zhao et al., 2005). Mounting evidence for p110a’s impor-
tance in cancer has spurred the development of PI3K-targeted
SIGNIFICANCE
Point mutations that block drug binding are likely to be a major mechanism of clinical resistance to PI3K-targeted cancertherapy. Here we report resistance mutations in the oncogenic PI3K isoform p110a, as well as a drug-sensitizing mutationthat will be useful for chemical genetic studies. This study anticipates p110a mutations that are likely to emerge againstPI3K-targeted drugs and identifies inhibitor classes that can overcome these resistance mutations. Our experiments inmammalian cells show that multitargeted inhibitors with additional PI3K pathway targets are less susceptible to drug resis-tance than selective PI3K inhibitors. The screening protocol described here is applicable to several other drug targets thatinhibit S. cerevisiae growth in addition to p110a.
Cancer Cell
PIK3CA Drug-Resistance Screen in S. cerevisiae
inhibitors, several of which are now entering clinical trials as can-
cer therapies (Marone et al., 2008; Ward and Finan, 2003).
As these targeted PI3K inhibitors progress through clinical tri-
als, it is likely that drug resistance will emerge in cancer patients
as it does to targeted protein kinase inhibitors (Daub et al., 2004).
The most frequent mechanism of resistance to protein kinase in-
hibitors in cancer therapy is point mutation of the target kinase
that blocks drug binding (Daub et al., 2004; Gorre et al., 2001;
Mahon et al., 2000; Shah et al., 2002; Shah and Sawyers,
2003), and mutation at one position in particular is frequently
observed across the protein kinase family, including BCR-ABL
(T315I), c-Kit (T670I), PDGFR (T674I), and EGFR (T790M) (Daub
et al., 2004). This position is termed the ‘‘gatekeeper’’ because
it controls access to a large hydrophobic pocket in which most
kinase inhibitors bind (Bishop et al., 2000; Liu et al., 1998).
Recent crystal structures of inhibitor-bound p110g have re-
vealed a similar large hydrophobic pocket, termed the affinity
pocket, which is occupied by the most potent PI3K inhibitors
(Knight et al., 2006). Structure-activity relationship (SAR) data
(Knight et al., 2006) and the crystal structure of p110a (Huang
et al., 2007) indicate that this affinity pocket is conserved in all
p110 isoforms, and structural alignment reveals that I848
occupies a position in p110a that is structurally analogous to
the gatekeeper residue in protein kinases (Walker et al., 1999).
This suggests the possibility that drug resistance in p110a might
arise as it often does in protein kinases, by mutation at the gate-
keeper residue I848 or at other residues lining the affinity pocket.
Although PI3Ks share significant sequence and structural
homology with the protein kinase family, the level of homology
is not high enough to confidently predict resistance mutations
in p110a from the corresponding mutations in protein kinases.
The identification and characterization of protein kinase resis-
tance mutations from cancer patients has led to the develop-
ment of effective second-generation inhibitors (Carter et al.,
2005; Quintas-Cardama et al., 2007; Shah et al., 2004). Remark-
ably, mutagenic screens in mammalian tissue culture are able to
reproduce these resistance mutations in vitro (Azam et al., 2003;
Engelman et al., 2006; von Bubnoff et al., 2005). These studies
are typically performed with some knowledge of resistance
mutations in the target kinase that have already been identified
in cancer patients. In this study, we sought to identify p110a re-
sistance mutations in vitro before they occur in cancer patients,
with no prior knowledge of clinical resistance in the PI3K family.
The structural diversity of PI3K inhibitors in clinical develop-
ment (Marone et al., 2008) suggests that a different spectrum
of resistance mutations may arise against each inhibitor. In order
to screen several PI3K inhibitors simultaneously, we used a PI3K
expression system in the budding yeast S. cerevisiae, which
allows an unlimited number of drug conditions to be screened
via replica plating or robotic pinning. Unlike previous mutagenic
screens in mammalian tissue culture, which rely on oncogenic
transformation by the target kinase (Azam et al., 2003; Engelman
et al., 2006; von Bubnoff et al., 2005), our screen is based on
growth inhibition induced by overexpression of membrane-
localized p110a (Rodriguez-Escudero et al., 2005; Tugendreich
et al., 2001). p110a activity severely inhibits growth in S. cerevi-
siae because it has no endogenous p110 homolog and therefore
does not produce or degrade endogenous PIP3 (Odorizzi et al.,
2000). Previous studies have shown that p110a-induced growth
inhibition can be partially reversed by the low-potency PI3K in-
hibitor LY294002 (Rodriguez-Escudero et al., 2005); in this study,
we used a structurally diverse panel of high-potency PI3K inhib-
itors that completely rescue S. cerevisiae growth. Drug-resistant
p110a mutants were identified by continued growth inhibition at
drug concentrations that rescue growth from wild-type p110a.
An unexpected benefit of the yeast screen was the ability to
identify drug sensitization in addition to drug resistance, by
enhanced growth rescue at low inhibitor concentrations that
do not rescue growth from wild-type p110a. Currently there is
no truly selective p110a inhibitor available, so the drug-sensi-
tized mutants that we identified will be valuable tools for the
study of p110a’s role in biological processes that rely on PI3K
signaling, from oncogenesis and tumor maintenance (Gupta
et al., 2007; Lim and Counter, 2005) to processes as varied as
neuronal development and animal behavior, viral replication,
and stem cell self-renewal (Cosker and Eickholt, 2007; Hale
and Randall, 2007; Kwon et al., 2006; Paling et al., 2004; Rod-
gers and Theibert, 2002). Additional benefits to screening with
yeast are the speed and simplicity of yeast culture in comparison
to mammalian tissue culture, the straightforward recovery of
plasmid DNA from screen hits for DNA sequencing, and the abil-
ity to screen inhibitors that would confound mammalian transfor-
mation assays by their off-target effects (Fan et al., 2006).
To determine whether p110a is susceptible to drug-resistant
mutations in a manner similar to protein kinases and to search
for drug-sensitized p110a mutants, we mutagenized residues
that line the affinity pocket and screened against a structurally
diverse panel of PI3K inhibitors. Using a high-throughput parallel
approach with robotic pinning, we generated a measure of cat-
alytic activity and an inhibition profile for every clone in our
mutant library, allowing for a detailed structure-function analysis
of the p110a affinity pocket. Potential drug-resistant and drug-
sensitizing mutations identified by the yeast screen were further
characterized in mammalian systems to rule out artifacts from
heterologous yeast expression by kinase assays following
expression in the human cell line HEK293T and by transforma-
tion assays with the human breast epithelial cell line MCF10A.
RESULTS
PI3K Activity Inhibits Growth in S. cerevisiae and CanBe Rescued by Selective PI3K InhibitorsIn order to use PI3K-induced toxicity to study p110a inhibition in
S. cerevisiae, we generated high-copy (2m) plasmids that
express p110a-CAAX under control of a GAL1 promoter and
transformed them into a wild-type strain, AFS92, and a drug-per-
meabilized strain, YRP1 (Derg6, Dpdr5, Dsnq2) (Gray et al.,
1998). When grown on galactose, both strains showed greater
than 1000-fold growth inhibition compared to empty vector
and kinase-dead controls (Figure 1A). PI3K-induced growth inhi-
bition was more severe in the YRP1 strain than in the AFS92
strain, and there was no difference in growth inhibition between
wild-type p110a-CAAX and the oncogenic H1047R mutant
(Figure 1A). The PI3K inhibitor PI-103 completely rescued growth
in the drug-permeabilized YRP1 strain but only partially rescued
growth in the wild-type AFS92 strain (Figure 1A). Growth rescue
by several other PI3K inhibitors was more efficient in YRP1 than
Cancer Cell 14, 180–192, August 12, 2008 ª2008 Elsevier Inc. 181
Cancer Cell
PIK3CA Drug-Resistance Screen in S. cerevisiae
Figure 1. Rescue of PI3K-Induced Growth Inhibition in S. cerevisiae by Selective PI3K Inhibitors
(A) Three-fold serial dilutions of AFS92 and YRP1 yeast strains containing the pURA3-2m-GAL1 plasmid expressing the indicated p110a-CAAX mutants were
spotted onto agar plates of SD �uracil media containing glucose, galactose, or galactose with 5 mM PI-103. All plates were incubated at 30�C for 2 days, except
for the YRP1 strains grown on galactose, which were incubated for 6 days.
(B) Reverse halo assays with YRP1-pURA3-2m-GAL1-p110a H1047R-CAAX grown on SD�uracil +galactose medium. Each plate was spotted with 10 ml of a 10
mM DMSO stock of PI3K inhibitor.
(C) Schematic overview of the S. cerevisiae growth inhibition screen used to identify drug-resistant and drug-sensitized p110a mutants.
in AFS92 (data not shown), and therefore the YRP1 strain was
chosen for use in all further experiments.
It was desirable to include as many diverse PI3K inhibitors as
possible in the planned screen of p110a mutants in order to
identify pan-inhibitor resistance mutations as well as inhibitor-
specific mutations and to give the best chance of identifying
a mutant/inhibitor pair that confers drug sensitivity. To test all
PI3K inhibitors for compatibility with the yeast assay format,
we used a variation on the traditional halo assay that we term
‘‘reverse halo assay.’’ In this assay, a PI3K inhibitor is spotted
182 Cancer Cell 14, 180–192, August 12, 2008 ª2008 Elsevier Inc.
onto a cellulose disc in the middle of a lawn of yeast, and instead
of inhibiting growth as in a traditional halo assay, the PI3K inhib-
itor rescues growth, creating a ‘‘growth halo’’ of healthy yeast.
The diameter and intensity of this halo depend on the inhibitor’s
potency, stability, diffusion rate, and lack of S. cerevisiae toxicity.
Two nonselective PI3K inhibitors, wortmannin and LY294002,
did not produce growth halos in the reverse halo assay
(Figure 1B). This is not surprising because wortmannin inhibits
the essential S. cerevisiae kinase STT4 at low nM concentrations
(Cutler et al., 1997), while LY294002 is only a mM inhibitor of
Cancer Cell
PIK3CA Drug-Resistance Screen in S. cerevisiae
Figure 2. p110a Gatekeeper Mutants Have Reduced Catalytic Activity
(A) Sequence alignment of p110a with protein kinases that display clinical resistance mutations at the gatekeeper position.
(B) ATP binding sites of Abl (PDB ID code 2G1T) and p110a (PDB ID code 2RD0). ATP from the p110g cocrystal (PDB ID code 1E8X) was overlaid onto the apo
p110a structure by structural alignment. The gatekeeper residue is colored red and labeled in both structures.
(C) Six-fold serial dilutions of YRP1-pURA3-2m-GAL1-p110a H1047R-CAAX strains with the indicated p110a mutations were spotted onto agar plates of
SD �uracil media containing either glucose or galactose. The plates were incubated at 30�C for 2 days (glucose) and 6 days (galactose).
(D) Distribution of residues at the gatekeeper position in the human PI3K and protein kinase families.
p110a and other PI3K family members (Brunn et al., 1996; Vlahos
et al., 1994). The imidopyrazine PIK-75 produced only a thin ring
of growth (Figure 1B), suggesting an exceedingly narrow dose
window in which a mutagenic screen could be performed. Five
structurally diverse p110a inhibitors, PIK-90, PIK-93, PI-103,
PW-12, and PP-110, produced growth halos of various size
and intensity (Figure 1B) and were therefore selected for use in
subsequent mutagenic screens.
Figure 1C illustrates how p110a-induced growth inhibition in
S. cerevisiae was used to screen for drug resistance and sensi-
tization. Residues of interest were subjected to saturation muta-
genesis with randomized primers, and the resulting libraries were
transformed into the permeabilized yeast strain YRP1. Individual
colonies were picked and arrayed into 384-pin format and then
replicated with a robotic pinner onto glucose- and galactose-
containing media to determine the relative colony size, and
therefore PI3K activity, of each mutant clone. Active mutants
were picked and arrayed onto new 384-pin format plates and
then replicated onto multiple PI3K inhibitor plates to screen for
drug resistance and sensitization.
Mutation of the p110a Gatekeeper Residue I848Initially we focused on the p110a gatekeeper residue, because in
protein kinases this position is the most frequent site of drug-re-
sistant mutations (Daub et al., 2004). Sequence alignment of
p110a with protein kinases revealed I848 to be its gatekeeper
residue (Figure 2A), which is positioned in the active site similarly
to the gatekeeper residue in protein kinases (Figure 2B). Using
site-directed mutagenesis with randomized primers, we mutated
I848 to every amino acid in the low-copy plasmid pURA3-CEN/
ARS-GAL1-p110a H1047R-CAAX. These plasmids were trans-
formed into YRP1, and the resulting strains were grown on either
glucose or galactose to determine the relative PI3K activity of
each mutant (Figure 2C). With the exception of mild growth inhi-
bition by the conservative mutations I848L, I848S, and I848V, the
other 16 p110a mutants did not inhibit yeast growth (Figure 2C),
suggesting that these mutants are catalytically inactive or
unstable when expressed in yeast. This result corroborates
and extends previous work showing loss of catalytic activity
with the I848A and I848G mutations (Alaimo et al., 2005). The
inability of p110a to tolerate nonconservative mutations at the
gatekeeper position is reflected in the evolutionary conservation
of lipid kinases at this position. Only isoleucine, leucine, methio-
nine, and valine are found at the gatekeeper position in the PI3K
family, and only isoleucine is found among the p110 isoforms,
while a greater diversity of amino acid side chains is observed
in the protein kinase family (Figure 2D).
A Mutagenic Screen of the p110a Affinity Pocketagainst a Diverse Panel of PI3K InhibitorsMinimal growth inhibition by p110a gatekeeper mutants in the
low-copy vector made screening for drug resistance impossible,
so the high-copy plasmid pURA3-2m-GAL1-p110a H1047R-
CAAX was used in all further experiments. I848 appeared
Cancer Cell 14, 180–192, August 12, 2008 ª2008 Elsevier Inc. 183
Cancer Cell
PIK3CA Drug-Resistance Screen in S. cerevisiae
Figure 3. Affinity Pocket Residues Chosen for Saturation Mutagenesis
(A) The p110a active site, with the adenine pocket colored blue-green and the affinity pocket colored pink. The p110a apo structure (PDB ID code 2RD0) was
overlaid with ATP (PDB ID code 1E8X) and PIK-90 (PDB ID code 2CHX) from p110g cocrystal structures.
(B) p110a affinity pocket residues chosen for saturation mutagenesis, shown on the p110a structure (PDB ID code 2RD0).
(C) Sequence alignment of the human PI3K family. Affinity pocket residues chosen for saturation mutagenesis are indicated by boxes.
relatively intolerant to mutation (Figure 2C), so the screen was
expanded to include seven additional residues surrounding the
p110a affinity pocket. These residues were chosen because
the affinity pocket is occupied by most potent PI3K inhibitors,
but not by ATP (Knight et al., 2006) (Figure 3A), making it a likely
site for drug-resistant mutations. I800, L807, L814, Y836, G837,
C838, and I848 were chosen based on proximity to the affinity
pocket and lack of interaction with the catalytic K802, DFG motif
(responsible for Mg2+ coordination), or ATP (Figure 3B). One ad-
ditional residue outside the affinity pocket, S854, was chosen
due to possible inhibitor-specific H bonds. These eight residues
are highly conserved in the PI3K family and almost 100% identi-
cal among the p110 isoforms (Figure 3C).
The chosen residues were mutagenized and screened against
PI3K inhibitors as described in Figure 1C. Each residue was sub-
jected to saturation mutagenesis with randomized NNK primers,
and the resulting eight libraries (see Figure S1 available online)
were transformed into the yeast strain YRP1. 384 colonies
from each NNK library were picked and arrayed onto 384-pin
format plates and then replicated onto glucose and galactose
plates to determine the PI3K activity of each clone by colony
size, measured with CellProfiler image analysis software
(Figure 4A). Active mutants with relative colony sizes less than
70% of wild-type were picked and arrayed onto new 384-pin
format plates and then replicated onto multiple PI3K inhibitor-
containing plates (Table S4) to screen for drug resistance and
sensitization. PIK-85 was used in place of PIK-90 for all screens;
it is identical to PIK-90 except for an enol group in place of
PIK-90’s amide. An array of pooled active clones replicated
184 Cancer Cell 14, 180–192, August 12, 2008 ª2008 Elsevier Inc.
onto either DMSO or 5 mM PI-103 is shown in Figure 4B. Potential
drug-resistant and drug-sensitized clones identified on the 384-
pin inhibitor plates were validated by serial dilution analysis
(Figure S2). Serial dilutions of the drug-resistant mutant strain
I800M and the drug-sensitized mutant strain L814C are shown
in Figure 4C.
Characterization of p110a Mutants by MammalianExpression and In Vitro Kinase AssayDrug-resistant and drug-sensitizing mutations identified by the
yeast screen require validation in a mammalian context to con-
firm that their altered activities and inhibitor sensitivities are not
due to membrane localization by the C-terminal CAAX box motif,
lack of a p85 binding partner, or other artifacts of S. cerevisiae
expression. Mutations of interest were introduced into Myc-
tagged p110a H1047R and then expressed in the human cell
line HEK293T for purification and enzymatic assay. Most of the
mutants tested had slightly reduced activity (Figure 5A): I800L
and I800M showed approximately 2-fold decreases in PI3K
activity, while L814C’s activity was significantly reduced. In vitro
IC50 values were determined for all of the p110a mutants against
the inhibitors used in the yeast screen, as well as BEZ-235, which
we obtained after the screen was performed. Mutation at I800
conferred the greatest drug resistance, with both I800L and
I800M showing approximately 5- to 10-fold decreases in inhibi-
tor potency for PIK-90, PIK-93, PI-103, and PP-110, although
I800L was sensitized to the compounds PW-12 and BEZ-235
(Figure 5B). Mutation at L814 conferred the greatest drug sensi-
tization, with the L814C mutant sensitized more than 10-fold to
Cancer Cell
PIK3CA Drug-Resistance Screen in S. cerevisiae
Figure 4. Growth Inhibition Screen of p110a Mutant Libraries in S. cerevisiae
(A) YRP1-pURA3-2m-GAL1-p110a H1047R-CAAX mutant library pinned onto SD�uracil +galactose and grown for 5 days at 30�C, and the distribution of colony
sizes for that plate relative to the same array plated on SD �uracil +glucose and grown for 3 days at 30�C (not shown).
(B) Representative image of pooled active clones from YRP1-pURA3-2m-GAL1-p110a H1047R-CAAX mutant libraries pinned onto SD �uracil +galactose with
either DMSO control or 5 mM PI-103 and grown for 5 days at 30�C, and the corresponding distributions of colony sizes relative to the same array plated on
SD �uracil +glucose and grown for 3 days at 30�C (not shown).
(C) Ten-fold serial dilutions of YRP1-pURA3-2m-GAL1-p110a H1047R-CAAX strains with the indicated p110a mutations spotted onto SD �uracil +galactose
media containing the indicated PI3K inhibitors and grown for 5 days at 30�C.
PIK-90 and PP-110 and approximately 100-fold to PIK-93
(Figure 5B).
I800L, I800M, and L814C Mutants Transform MCF10ACells and Confer Resistance or Sensitizationto PI3K InhibitorsThe oncogenic potential of the most drug-resistant and drug-
sensitized p110a mutants was assessed in the untransformed
breast cell line MCF10A, which requires epidermal growth factor
(EGF) for growth under normal tissue culture conditions but can
be transformed to EGF-independent growth by p110a H1047R
(Isakoff et al., 2005). I800L, I800M, L814C, wild-type, and
kinase-dead (K802R) mutations were made in the retroviral
plasmid pMIG-p110a H1047R and then transduced into
MCF10A cells. The pMIG vector contains an internal ribosome
entry site (IRES)-GFP sequence: infected MCF10A cell lines
were all sorted by fluorescence-activated cell sorting (FACS) to
greater than 95% GFP positive for use in further experiments.
p110a-expressing MCF10A cell lines were cultured in medium
without EGF to determine whether the drug-resistant and drug-
sensitized mutant p110a H1047Rs could support EGF-indepen-
dent growth and to monitor changes in PI3K inhibitor sensitivity
for each mutant. The I800L and I800M mutant cell lines grew at
the same rate as p110a H1047R, while the L814C mutation
caused a small but consistent decrease in growth (Figure 6A).
I800L and I800M conferred resistance to all inhibitors, with the
exception that I800L was sensitized to PW-12 and BEZ-235
(Figure 6B). PI3K inhibitor resistance was most pronounced
with the selective PI3K inhibitors PIK-90 and PIK-93 and was
less dramatic against the multitargeted PI3K inhibitors PI-103
and BEZ-235 (mTOR), PW-12 (multiple protein kinases), and
PP-110 (receptor tyrosine kinases) (Figure 6B) (Fan et al., 2006;
Hayakawa et al., 2007a, 2007b; Knight et al., 2006; Raynaud
et al., 2007; Stauffer et al., 2008; B.A. and K.M.S., unpublished
data). Similarly, L814C conferred strong sensitization of cell
growth to PIK-90 and PIK-93, moderate sensitization to PW-12
and PP-110, and minimal or no sensitization to PI-103 and
BEZ-235 (Figure 6C), again likely due to the selectivity of PIK-
90 and PIK-93 and the relative promiscuity of PW-12, PP-110,
PI-103, and BEZ-235. As a reference, inhibitory values for the
PI3K inhibitors against p110a and selected off-target protein
kinases are shown in Table S1.
I800L, I800M, and L814C Mutants Retain the Abilityto Induce EGF-Independent Akt Phosphorylationin MCF10A CellsPhosphorylation of Akt residues T308 and S473 was monitored
in the I800L-, I800M-, and L814C-expressing MCF10A cell lines
to determine whether these p110a mutants retain the ability to
activate the canonical downstream PI3K signaling pathway.
The I800L and I800M mutations to p110a H1047R produced
high phospho-Akt levels comparable to p110a H1047R after
24 hr EGF starvation, but L814C produced reduced, although
still detectable, levels (Figure 7A). Phospho-Akt could be fully re-
covered in all cell lines by 1 hr treatment with EGF-containing
medium, and this stimulation could be blocked in all lines by
30 mM PI-103 (Figure 7A). GFP levels, which are coupled to
p110a expression by an IRES promoter, were greatly reduced
Cancer Cell 14, 180–192, August 12, 2008 ª2008 Elsevier Inc. 185
Cancer Cell
PIK3CA Drug-Resistance Screen in S. cerevisiae
Figure 5. Characterization of Yeast Screen Hits by Mammalian Expression and In Vitro Kinase Assay
(A) The indicated Myc-tagged p110a H1047R mutants were immunoprecipitated from HEK293T cells and assayed for PI3K activity. Data are represented as
mean ± SEM.
(B) In vitro IC50 values were determined for each p110a mutant against the indicated PI3K inhibitors at 10 mM ATP. Changes in inhibitor sensitivity are shown as the
ratio of mutant IC50/wild-type (WT) IC50. ‘‘ND’’ (not determined) indicates that the data from the performed dose-response experiment were insufficient to
generate an IC50 curve.
in all p110a-expressing cell lines, and p110a levels were not
highly elevated in comparison to the empty vector control
(Figure 7A). This suggests that MCF10A cells cannot tolerate
overexpression of p110a H1047R and shows that expression
near endogenous levels is sufficient for Akt activation and trans-
formation to EGF independence.
The I800L, I800M, and L814C cell lines were treated with PI3K
inhibitors in serial dilution to determine how each mutation affects
the inhibitor sensitivity of phospho-Akt levels. Similar to the
MCF10A growth curves, the most striking resistance and sensiti-
zation occurred with the selective PI3K inhibitors PIK-90 and
PIK-93, most likely because these results were not confounded
by off-target effects. I800L was approximately 3-fold resistant
to PIK-90 and PIK-93, I800M was approximately 10-fold resistant
to PIK-93, and L814C was approximately 5-fold and 30-fold sen-
sitized to PIK-90 and PIK-93 respectively (Figure 7B), consistent
with in vitro kinase assays and the MCF10A growth curves. Sig-
nificant resistance was not observed against the multitargeted
inhibitors PI-103, PW-12, PP-110, and BEZ-235, but sensitization
was seen with I800L to PW-12 and L814C to PW-12, PP-110, and
BEZ-235 (Figure 7B), again consistent with in vitro kinase assays
and the MCF10A growth curves. Slight resistance and sensitiza-
tion were observed with the I800L mutant to the dual PI3K/mTOR
inhibitors PI-103 and BEZ-235, respectively. This trend was seen
186 Cancer Cell 14, 180–192, August 12, 2008 ª2008 Elsevier Inc.
in phospho-T308 Akt but not phospho-S473, most likely because
mTOR is the kinase that phosphorylates S473 in these cells (Sar-
bassov et al., 2005).
Tolerance to Mutation in the p110a Affinity PocketIn order to determine how tolerant each p110a affinity pocket res-
idue is to mutation, we compared the colony size distributions
generated for each residue in the course of our mutagenic screen,
as seen in Figure 4A. All mutant libraries displayed a peak of
relative colony sizes close to 1.0, corresponding to no growth
inhibition and therefore kinase-dead mutants (Figure 8A). The
majority of mutant libraries also displayed a second peak of
smaller colony sizes, corresponding to growth inhibition and
therefore PI3K activity (Figure 8A). The tolerance to mutation for
each residue was quantified by S(1� x)2, where x equals relative
colony size. The resulting values are converted into heat map
color scale and displayed on p110a in Figure 8B. The I800,
I807, L814, and G837 residues are most tolerant to mutation;
C837, I848, and S854 are less tolerant to mutation; and Y836
appears to be completely intolerant to mutation, with even the
conservative mutation Y836F causing a substantial loss of kinase
activity (Figure S2). Intolerance to mutation at Y836 corroborates
previous results with p110a (Alaimo et al., 2005). Recently pub-
lished work has shown that the mutation corresponding to
Cancer Cell
PIK3CA Drug-Resistance Screen in S. cerevisiae
Y836M in PI4KIIIb produces an active kinase that is resistant to
wortmannin and PIK-93 and that the same mutation in PI4KIIIa
kills all catalytic activity (Balla et al., 2008). It is currently unclear
why this mutation is tolerated in PI4KIIIb but abolishes kinase ac-
tivity in PI4KIIIa and p110a. Further experiments with additional
PI3K family members may better reveal the sequence determi-
nants for mutation tolerance and inhibitor sensitivity.
DISCUSSION
We describe here the development of a resistance screen based
on PI3K-induced growth inhibition of S. cerevisiae. Unlike resis-
tance screens based on oncogenic transformation of mamma-
lian cells, this screen allows for the identification of drug-sensi-
tized mutants in addition to drug-resistant mutants and allows
for detailed structure-function analysis of the drug target,
because activity levels and inhibitor sensitivities are determined
for every clone in the mutant library. Previous studies have used
growth inhibition in S. cerevisiae by heterologous expression of
a mammalian drug target to screen a compound library for po-
tent inhibitors (Boschelli et al., 2001), but to our knowledge,
such a screen has not been used to identify drug-resistant and
Figure 6. Yeast Screen Hits I800L, I800M,
and L814C Confer EGF-Independent
Growth to MCF10A Cells and Maintain
Altered Inhibitor Sensitivities In Vivo
(A) MCF10A cell lines expressing the indicated
p110a mutants were cultured in growth medium
lacking EGF and monitored for growth by alamar
blue assay.
(B and C) Growth of the I800L, I800M, and L814C
MCF10A cell lines in medium lacking EGF was
monitored as in (A) in the presence of the indicated
PI3K inhibitors.
Data are represented as mean ± SEM.
drug-sensitizing mutations in the target
protein. In addition to p110a, several
other drug targets inhibit S. cerevisiae
growth, including Akt1, PDK1, p38 ki-
nase, Src, RhoA, RhoC, PARP-1, and
HIV protease (Brugge et al., 1987; Tugen-
dreich et al., 2001), and the methods
described here should be broadly appli-
cable for the identification of drug-resis-
tant and drug-sensitizing mutations in
these proteins as well.
Our results in S. cerevisiae indicate that
the p110a gatekeeper residue I848 is not
a hotspot for inhibitor resistance in p110a
as it is in the protein kinase family (Fig-
ure 2C; Figure 8A). While protein kinases
can tolerate dramatic mutations at the
gatekeeper position, from smaller,
space-creating mutations that sensitize
protein kinases to analog-specific inhibi-
tors (Bishop et al., 2000) to larger, steric
clash-inducing mutations that confer in-
hibitor resistance (Daub et al., 2004), it appears that p110a can-
not tolerate significant mutations at this position. This suggests
the possibility that the PI3K fold is less tolerant of mutations
than the protein kinase fold in this region, which may reduce
the clinical development of resistance to p110a-targeted drugs,
although a more complete mutagenic screen of p110a is re-
quired to prove this conclusively.
In contrast to I848, the residue I800 is tolerant to mutations
(Figure 8A), two of which confer drug resistance: I800L and
I800M. The resistance conferred by these mutations is smaller
than that observed for the T315I gatekeeper mutant of BCR-ABL
but comparable to most other mutations that confer clinical resis-
tance to imatinib (Shah et al., 2002; von Bubnoff et al., 2005). This
residue is a potential hotspot for clinical resistance to PI3K inhib-
itors; fortunately, our screen identified PW-12 and BEZ-235 as
inhibitors that potently target the I800L mutant. The fact that
I800L confers resistance to all other inhibitors but sensitizes
p110a to PW-12 and BEZ-235 suggests that these inhibitors
share a similar binding mode that differs from other scaffolds
near residue I800. There are several inhibitor-bound PI3K struc-
tures available, but the cocrystal structures of PW-12 and BEZ-
235 have not been reported, so it is difficult to rationalize these
Cancer Cell 14, 180–192, August 12, 2008 ª2008 Elsevier Inc. 187
Cancer Cell
PIK3CA Drug-Resistance Screen in S. cerevisiae
trends. Interestingly, the residue corresponding to I800 in mTOR
is also leucine, and mTOR is potently targeted by BEZ-235.
The drug-sensitizing mutation L814C will be a valuable tool to
study the effect of p110a inhibition in various biological systems,
because there is no selective p110a inhibitor currently available.
One possible concern with the L814C mutation is that its ‘‘sensi-
tizing’’ effect may simply be due to the loss of enzymatic activity
(Figure 5A), but in vitro kinase assays reveal large shifts in IC50
values (Figure 5B), and yeast serial dilution analysis (Figure 4C)
and MCF10A transformation assays (Figure 6A) indicate that
this mutant behaves similarly to wild-type in a cellular context,
although it does show reduced Akt phosphorylation (Figure 7B).
Transferring the L814C, I800L, and I800M mutations to other
members of the PI3K family presents an excellent opportunity to
study the cellular roles of each p110 isoform by specific inhibition
or drug resistance. I800 and L814 are absolutely conserved
Figure 7. Yeast Screen Hits I800L, I800M,
and L814C Produce EGF-Independent Akt
Phosphorylation in MCF10A Cells, which
Maintains Altered Inhibitor Sensitivities
(A) MCF10A cell lines expressing the indicated
p110a mutants were cultured in growth medium
lacking EGF for 24 hr and then treated with the in-
dicated combinations of normal growth medium
(GM) and 30 mM PI-103. After 1 hr, the cells were
lysed and subjected to western blot analysis with
the indicated antibodies.
(B) The indicated MCF10A cell lines were cultured
in growth medium lacking EGF for 24 hr and then
treated for 1 hr with serial dilutions of the indicated
PI3K inhibitors, after which the cells were lysed
and subjected to western blot analysis with the in-
dicated antibodies. PI3K inhibitor concentrations
are indicated in mM.
among the p110 isoforms and highly con-
served among the rest of the PI3K family,
so the likelihood of successfully generat-
ing drug-resistant and drug-sensitized
mutants for each family member appears
high. Concerning the loss of activity
caused by gatekeeper mutations in
p110a, it will be interesting to test whether
this trend holds for other PI3K family mem-
bers or is unique to p110a.
The advent of high-throughput screen-
ing and the great success of imatinib led
the pharmaceutical industry to focus on
highly specific kinase inhibitors, but cur-
rently a multitargeted approach is gaining
acceptance (Branca, 2005; Jimeno and
Hidalgo, 2006). Inhibiting multiple targets
can increase efficacy (Fan et al., 2006)
and theoretically should decrease the
likelihood of drug resistance, although
no study to our knowledge has conclu-
sively shown a reduced likelihood of clin-
ical resistance with multitargeted versus
highly specific drugs. Our results with
the MCF10A cell line (Figure 6; Figure 7; Figure S3) show that
multitargeted inhibitors are not as susceptible to drug resistance
by mutation of the single target p110a, especially when they tar-
get additional kinases in the PI3K signaling pathway.
PI-103 and BEZ-235 block PI3K signaling downstream of
p110a by inhibiting mTOR, and PP-110 blocks PI3K signaling
upstream of p110a by inhibiting RTKs. The ability of these three
inhibitors to block p110a H1047R-dependent MCF10A growth is
largely unaffected by p110a resistance mutations, suggesting
that mutation of additional targets is required to confer drug
resistance. The possibility of accumulating multiple resistance
mutations in a single cancer cell seems unlikely, but slight growth
advantages at each step may increase the odds of successive
mutations, especially during prolonged treatment. The rate and
probability of resistance mutation accumulation will be an impor-
tant area of study for multitargeted drugs in cancer therapy,
188 Cancer Cell 14, 180–192, August 12, 2008 ª2008 Elsevier Inc.
Cancer Cell
PIK3CA Drug-Resistance Screen in S. cerevisiae
Figure 8. Tolerance to Mutation in the p110a Affinity Pocket
(A) Colony size distributions for 384 colony arrays of the indicated p110a mutant libraries, grown in the YRP1 strain on SD �uracil +galactose medium as
described in Figure 4A.
(B) Tolerance to mutation as calculated by S(1� x)2 from the distributions in (A), where x equals relative colony size. These values were converted into heat map
values and are shown on the p110a crystal structure (PDB ID code 2RD0) with the PI3K inhibitor PIK-90 from the p110g cocrystal structure (PDB ID code 2CHX)
overlaid by structural alignment.
accessible by careful monitoring of clinical trials as well as mam-
malian tissue culture screens.
By focusing on a small subset of p110a residues that line the
affinity pocket, we have identified a potential resistance hotspot
at I800 that confers 5- to 30-fold resistance to most PI3K inhib-
itors. While this discovery should help guide the design of sec-
ond-generation PI3K inhibitors, a more complete mutagenic
screen of p110a will be necessary to uncover the full spectrum
of resistance mutations.
EXPERIMENTAL PROCEDURES
Plasmids
Yeast Expression Plasmids
A C-terminal myristoylation sequence (CAAX box) was added to human p110a
H1047R by three rounds of PCR amplification with the primers (1) hp110a-FM
and hp110a-CRM and (2) hp110a-FL and hp110a-CRL. (All oligonucleotide
primers used for cloning are described in Table S2.) The resulting fragment
was cloned into a high-copy (2m) URA3 yeast expression vector with a GAL1
promoter by gap repair/homologous recombination to create the vector
pURA3-2m-GAL1-p110a H1047R-CAAX. Wild-type and kinase-dead plasmids
were created by site-directed mutagenesis with the primers hp110a
R1047H-F, hp110a R1047H-R, hp110a K802R-F, and hp110a K802R-R.
Mammalian Expression Plasmids
An N-terminal Myc tag was added to human p110a H1047R by PCR with
the primers p110a-BamHI-NtermMyc-F and p110a-EcoRV-R. The resulting
fragment was digested with BamHI and EcoRV and then ligated into the
mammalian expression vector pcDNA3 to create pcDNA3-Myc-p110a H1047R.
Several point mutations to this vector were made by site-directed mutagenesis.
Retroviral Plasmids
Human p110a H1047R was PCR amplified with the primers hp110aF-XhoI and
hp110aR-HpaI, and the resulting fragment was digested with XhoI and HpaI
and then ligated into the IRES-GFP retroviral vector pMIG to create pMIG-
p110a H1047R. Several p110a point mutations to this vector were made by
site-directed mutagenesis.
Murine Ecotropic Pseudotyping Vector
pcDNA3-EcoR plasmid DNA encoding the murine ecotropic receptor EcoR/
MCAT-1 was a generous gift from Jeffrey Henise.
Cancer Cell 14, 180–192, August 12, 2008 ª2008 Elsevier Inc. 189
Cancer Cell
PIK3CA Drug-Resistance Screen in S. cerevisiae
Yeast Strains and Media
The S. cerevisiae strains YRP1 (Derg6, Dpdr5, Dsnq2) and AFS92 were used
for all experiments. Strains were grown at 30�C on SD �uracil +glucose or +
galactose. PI3K inhibitors were added to media in DMSO at 1:100. YRP1
was transformed with plasmid DNA as described in Supplemental Data.
Reverse Halo Assay
Log-phase YRP1-pURA3-2m-GAL1-p110a H1047R-CAAX cultures grown in
SD �uracil +glucose were washed three times with water and then spread
into a lawn on SD�uracil +galactose agarose plates at approximately 106 cells
per plate. After drying, a small cellulose disc was placed in the middle of each
plate, and 10 ml of a DMSO inhibitor stock was then spotted onto the cellulose
disc. The plates were incubated at 30�C for 5–7 days before imaging.
PI3K Inhibitors
BEZ-235 was a generous gift from Yi Liu. All other inhibitors used in this study
were synthesized following previously reported protocols (Knight et al., 2006;
Stauffer et al., 2008; B.A. and K.M.S., unpublished data). In all tissue culture
experiments, DMSO inhibitor stocks were used at 1:1000.
Library Construction
The yeast expression vector pURA3-2m-GAL1-p110a H1047R-CAAX was mu-
tagenized at the residues I800, L807, L814, Y836, G837, C838, and S854 by
QuikChange PCR with degenerate NNK primers (Table S3), where N = 25%
A, 25% C, 25% G, 25% T and K = 50% G, 50% T. The resulting PCR reactions
were purified with a QIAGEN PCR cleanup kit, DpnI digested for 90 min, and
then repurified. Five microliters of each purified, digested PCR product was
transformed into TOP-10 One Shot chemically competent E. coli (Invitrogen)
and plated onto a single 10 cm plate of Luria broth medium with carbenicillin
antibiotic, yielding 1–4 3 103 colonies per transformation. Colonies were
grown for 2 days at 37�C and then pooled by scraping, transferred to 1.5 ml
tubes, and spun down to pellets of approximately 0.5 ml. Plasmid DNA was
isolated from each pellet with a QIAGEN miniprep kit and verified by restriction
digest and DNA sequencing (Figure S1).
Screening and Image Analysis
p110a mutant libraries were transformed into YRP1 by electroporation and
plated onto SD �uracil +glucose medium. 384 colonies from each library
were arrayed by hand and then replicated with a Virtek colony arrayer to obtain
uniformly sized colonies. The arrays were further replicated onto two media
conditions: SD �uracil +galactose medium with added PI3K inhibitor or
DMSO alone, and SD �uracil +glucose medium. The plates were incubated
at 30�C until the colonies had grown sufficiently (2–7 days, depending on strain
and media conditions) and then photographed. Colony size was calcu-
lated with CellProfiler image analysis software (http://www.cellprofiler.org/),
and each SD �uracil +galactose colony size value was divided by the corre-
sponding SD �uracil +glucose colony size to normalize for variation in pinning
efficiency.
In Vitro PI3K Assays
pcDNA3-p110a plasmid DNA was transfected into HEK293T cells with Lipo-
fectamine 2000 (Invitrogen). After 48 hr, the cells were trypsinized, washed
with PBS, and pelleted for storage at �80�C. Pellets were lysed by vortexing
in PI3K lysis buffer (50 mM Tris [pH 7.4], 300 mM NaCl, 5 mM EDTA, 0.02%
NaN3, 1% Triton X-100, protease inhibitor cocktail tablets [Roche], 8 mM so-
dium orthovanadate, 83 mM PMSF, 13 phosphatase inhibitor cocktails 1 and 2
[Sigma]) and then immunoprecipitated by overnight incubation with Anti-
c-Myc Agarose Affinity Gel (Sigma-Aldrich). The immunoprecipitates were
washed twice with buffer A (PBS, 1 mM EDTA, 1% Triton X-100), twice with
buffer B (100 mM Tris [pH 7.4], 500 mM LiCl, 1 mM EDTA), twice with buffer
C (50 mM Tris [pH 7.4], 100 mM NaCl), twice with PBS, and then assayed
for PI3K activity in 96-well format essentially as described previously (Knight
et al., 2007). Briefly, immunoprecipitated Myc-p110a was incubated ‘‘on
bead’’ with shaking at 25�C with 100 mg/ml phosphatidylinositol, 10 mCi
[g-32P]ATP, 10 mM ATP, 1 mg/ml BSA, 25 mM HEPES (pH 7.4), 10 mM
MgCl2, 1:50 DMSO with or without PI3K inhibitor. After 1 hr, 4 ml of each reac-
tion was spotted onto a dry nitrocellulose membrane prerinsed with wash
solution (1 M NaCl, 1% H3PO4). After the spots dried, the membrane was
190 Cancer Cell 14, 180–192, August 12, 2008 ª2008 Elsevier Inc.
washed five times with wash solution, dried with a heat lamp, and exposed
on a phosphor screen overnight. The phosphor screen was then scanned
with a Typhoon phosphorimager (GE Healthcare), and the resulting data
were quantified with the MATLAB script Spot (Knight et al., 2007).
Mammalian Cell Lines and Cell Culture
MCF10A cells (American Type Culture Collection) were cultured at 37�C, 5%
CO2 in MCF10A growth medium (1:1 DMEM:F12 supplemented with 5% fil-
tered heat-inactivated horse serum, 20 ng/ml EGF, 100 mg/ml hydrocortisone,
1 ng/ml cholera toxin, 10 mg/ml insulin, and penicillin/streptomycin) as de-
scribed previously (Debnath et al., 2003).
Generation of p110a-Expressing MCF10A Cell Lines
Mutant p110a-expressing MCF10A lines were created by retroviral infection.
Ecotropic p110a viral stocks were generated by transfecting pMIG-p110a
plasmid DNA into the Phoenix Eco cell line. Retroviral supernatants were col-
lected at 48, 72, 96, and 120 hr; spun at 500 rpm for 5 min at 4�C; and stored in
0.5 ml aliquots at �80�C.
Human MCF10A cells were pseudotyped for infection with murine ecotropic
virus by transient transfection with pcDNA3-EcoR/MCAT1. Transfection was
performed by nucleofection (Amaxa Nucleofector) following the manufac-
turer’s instructions, and the transfected cells were plated into six-well plates.
Twenty-four hours after nucleofection with pcDNA3-EcoR/MCAT1, the
MCF10A cells were infected with thawed p110a viral stocks for 12 hr, switched
to growth medium for 6 hr, and then expanded into 75 cm2 flasks. After 3 days,
the infected cells were FACS sorted for GFP expression.
MCF10A Western Blots
40%–70% confluent MCF10A cultures were starved for 24 hr with MCF10A
medium lacking EGF and insulin and then treated for 60 min with PI3K inhibi-
tors or DMSO control. After 60 min, the cells were lysed with PI3K lysis buffer
and subjected to western blotting with antibodies purchased from Santa Cruz
(anti-GFP) and Cell Signaling (all other antibodies) as directed by the supplier.
MCF10A Transformation Assays
MCF10A cell lines were assayed for oncogenic transformation by their ability
to grow in medium lacking EGF. Cells were seeded into black-sided, clear-
bottom 96-well plates at 2 3 103 cells per well in 100 ml growth medium. After
2 days, the cells were washed with PBS and then switched to 100 ml growth
medium lacking EGF. Cell proliferation was monitored every 2 days by incuba-
tion with resazurin (alamar blue). A stock solution of 120 mg/ml resazurin in PBS
was added to the cells at 1:20, incubated for 2 hr at 37�C and for another 15
min at room temperature, and then assayed for fluorescence with an excitation
wavelength of 520 nm and an emission wavelength of 590 nm. After fluores-
cence measurement, the resazurin-containing medium was replaced with
100 ml fresh growth medium lacking EGF, and the cells were returned to
37�C, 5% CO2.
Structural Analysis
Visualization and structural alignment of X-ray crystal structures was
performed with the PyMOL molecular graphics system (http://pymol.
sourceforge.net/).
SUPPLEMENTAL DATA
The Supplemental Data include Supplemental Experimental Procedures, Sup-
plemental References, three figures, and four tables and can be found with this
article online at http://www.cancercell.org/cgi/content/full/14/2/180/DC1/.
ACKNOWLEDGMENTS
We are grateful to Karl Kulcher for the YRP1 strain; Yi Liu at Intellikine for the
inhibitor BEZ-235; Dorothea Fiedler, Assen Roguev, and Nevan Krogan for as-
sistance with the colony arrayer; Sohye Kang and Peter Vogt for p110a
H1047R-containing plasmids; Brandon Tavshanjian for assistance with gener-
ation of the MCF10A cell lines, and Jeffrey Henise and Jack Taunton for the
pcDNA3-EcoR/MCAT1 vector. This work was supported by the Howard
Hughes Medical Institute.
Cancer Cell
PIK3CA Drug-Resistance Screen in S. cerevisiae
Received: April 10, 2008
Revised: May 19, 2008
Accepted: June 25, 2008
Published: August 11, 2008
REFERENCES
Alaimo, P.J., Knight, Z.A., and Shokat, K.M. (2005). Targeting the gatekeeper
residue in phosphoinositide 3-kinases. Bioorg. Med. Chem. 13, 2825–2836.
Azam, M., Latek, R.R., and Daley, G.Q. (2003). Mechanisms of autoinhibition
and STI-571/imatinib resistance revealed by mutagenesis of BCR-ABL. Cell
112, 831–843.
Balla, A., Tuymetova, G., Toth, B., Szentpetery, Z., Zhao, X., Knight, Z.A., Sho-
kat, K., Steinbach, P.J., and Balla, T. (2008). Design of drug-resistant alleles of
type-III phosphatidylinositol 4-kinases using mutagenesis and molecular
modeling. Biochemistry 47, 1599–1607.
Bishop, A.C., Ubersax, J.A., Petsch, D.T., Matheos, D.P., Gray, N.S., Blethrow,
J., Shimizu, E., Tsien, J.Z., Schultz, P.G., Rose, M.D., et al. (2000). A chemical
switch for inhibitor-sensitive alleles of any protein kinase. Nature 407, 395–
401.
Boschelli, D.H., Wang, Y.D., Ye, F., Wu, B., Zhang, N., Dutia, M., Powell, D.W.,
Wissner, A., Arndt, K., Weber, J.M., and Boschelli, F. (2001). Synthesis and Src
kinase inhibitory activity of a series of 4-phenylamino-3-quinolinecarbonitriles.
J. Med. Chem. 44, 822–833.
Branca, M.A. (2005). Multi-kinase inhibitors create buzz at ASCO. Nat. Bio-
technol. 23, 639.
Brugge, J., Hung, M.C., and Mills, G.B. (2007). A new mutational AKTivation in
the PI3K pathway. Cancer Cell 12, 104–107.
Brugge, J.S., Jarosik, G., Andersen, J., Queral-Lustig, A., Fedor-Chaiken,
M., and Broach, J.R. (1987). Expression of Rous sarcoma virus transforming
protein pp60v-src in Saccharomyces cerevisiae cells. Mol. Cell. Biol. 7,
2180–2187.
Brunn, G.J., Williams, J., Sabers, C., Wiederrecht, G., Lawrence, J.C., Jr., and
Abraham, R.T. (1996). Direct inhibition of the signaling functions of the mam-
malian target of rapamycin by the phosphoinositide 3-kinase inhibitors, wort-
mannin and LY294002. EMBO J. 15, 5256–5267.
Cantley, L.C. (2002). The phosphoinositide 3-kinase pathway. Science 296,
1655–1657.
Carter, T.A., Wodicka, L.M., Shah, N.P., Velasco, A.M., Fabian, M.A., Treiber,
D.K., Milanov, Z.V., Atteridge, C.E., Biggs, W.H., 3rd, Edeen, P.T., et al. (2005).
Inhibition of drug-resistant mutants of ABL, KIT, and EGF receptor kinases.
Proc. Natl. Acad. Sci. USA 102, 11011–11016.
Chang, H.W., Aoki, M., Fruman, D., Auger, K.R., Bellacosa, A., Tsichlis, P.N.,
Cantley, L.C., Roberts, T.M., and Vogt, P.K. (1997). Transformation of chicken
cells by the gene encoding the catalytic subunit of PI 3-kinase. Science 276,
1848–1850.
Cosker, K.E., and Eickholt, B.J. (2007). Phosphoinositide 3-kinase signalling
events controlling axonal morphogenesis. Biochem. Soc. Trans. 35, 207–210.
Cully, M., You, H., Levine, A.J., and Mak, T.W. (2006). Beyond PTEN muta-
tions: the PI3K pathway as an integrator of multiple inputs during tumorigene-
sis. Nat. Rev. Cancer 6, 184–192.
Cutler, N.S., Heitman, J., and Cardenas, M.E. (1997). STT4 is an essential
phosphatidylinositol 4-kinase that is a target of wortmannin in Saccharomyces
cerevisiae. J. Biol. Chem. 272, 27671–27677.
Daub, H., Specht, K., and Ullrich, A. (2004). Strategies to overcome resistance
to targeted protein kinase inhibitors. Nat. Rev. Drug Discov. 3, 1001–1010.
Debnath, J., Muthuswamy, S.K., and Brugge, J.S. (2003). Morphogenesis and
oncogenesis of MCF-10A mammary epithelial acini grown in three-dimen-
sional basement membrane cultures. Methods 30, 256–268.
Engelman, J.A., Mukohara, T., Zejnullahu, K., Lifshits, E., Borras, A.M., Gale,
C.M., Naumov, G.N., Yeap, B.Y., Jarrell, E., Sun, J., et al. (2006). Allelic dilution
obscures detection of a biologically significant resistance mutation in EGFR-
amplified lung cancer. J. Clin. Invest. 116, 2695–2706.
Fan, Q.W., Knight, Z.A., Goldenberg, D.D., Yu, W., Mostov, K.E., Stokoe, D.,
Shokat, K.M., and Weiss, W.A. (2006). A dual PI3 kinase/mTOR inhibitor
reveals emergent efficacy in glioma. Cancer Cell 9, 341–349.
Fruman, D.A., Meyers, R.E., and Cantley, L.C. (1998). Phosphoinositide
kinases. Annu. Rev. Biochem. 67, 481–507.
Gorre, M.E., Mohammed, M., Ellwood, K., Hsu, N., Paquette, R., Rao, P.N.,
and Sawyers, C.L. (2001). Clinical resistance to STI-571 cancer therapy
caused by BCR-ABL gene mutation or amplification. Science 293, 876–880.
Gray, N.S., Wodicka, L., Thunnissen, A.M., Norman, T.C., Kwon, S., Espinoza,
F.H., Morgan, D.O., Barnes, G., LeClerc, S., Meijer, L., et al. (1998). Exploiting
chemical libraries, structure, and genomics in the search for kinase inhibitors.
Science 281, 533–538.
Gupta, S., Ramjaun, A.R., Haiko, P., Wang, Y., Warne, P.H., Nicke, B., Nye, E.,
Stamp, G., Alitalo, K., and Downward, J. (2007). Binding of ras to phosphoino-
sitide 3-kinase p110alpha is required for ras-driven tumorigenesis in mice. Cell
129, 957–968.
Hale, B.G., and Randall, R.E. (2007). PI3K signalling during influenza A virus
infections. Biochem. Soc. Trans. 35, 186–187.
Hayakawa, M., Kaizawa, H., Kawaguchi, K., Ishikawa, N., Koizumi, T., Ohishi,
T., Yamano, M., Okada, M., Ohta, M., Tsukamoto, S., et al. (2007a). Synthesis
and biological evaluation of imidazo[1,2-a]pyridine derivatives as novel PI3
kinase p110alpha inhibitors. Bioorg. Med. Chem. 15, 403–412.
Hayakawa, M., Kaizawa, H., Moritomo, H., Koizumi, T., Ohishi, T., Yamano, M.,
Okada, M., Ohta, M., Tsukamoto, S., Raynaud, F.I., et al. (2007b). Synthesis
and biological evaluation of pyrido[30,20:4,5]furo[3,2-d]pyrimidine derivatives
as novel PI3 kinase p110alpha inhibitors. Bioorg. Med. Chem. Lett. 17,
2438–2442.
Hennessy, B.T., Smith, D.L., Ram, P.T., Lu, Y., and Mills, G.B. (2005). Exploit-
ing the PI3K/AKT pathway for cancer drug discovery. Nat. Rev. Drug Discov. 4,
988–1004.
Huang, C.H., Mandelker, D., Schmidt-Kittler, O., Samuels, Y., Velculescu, V.E.,
Kinzler, K.W., Vogelstein, B., Gabelli, S.B., and Amzel, L.M. (2007). The struc-
ture of a human p110alpha/p85alpha complex elucidates the effects of onco-
genic PI3Kalpha mutations. Science 318, 1744–1748.
Ikenoue, T., Kanai, F., Hikiba, Y., Obata, T., Tanaka, Y., Imamura, J., Ohta, M.,
Jazag, A., Guleng, B., Tateishi, K., et al. (2005). Functional analysis of PIK3CA
gene mutations in human colorectal cancer. Cancer Res. 65, 4562–4567.
Isakoff, S.J., Engelman, J.A., Irie, H.Y., Luo, J., Brachmann, S.M., Pearline,
R.V., Cantley, L.C., and Brugge, J.S. (2005). Breast cancer-associated PIK3CA
mutations are oncogenic in mammary epithelial cells. Cancer Res. 65, 10992–
11000.
Jimeno, A., and Hidalgo, M. (2006). Multitargeted therapy: can promiscuity
be praised in an era of political correctness? Crit. Rev. Oncol. Hematol.
59, 150–158.
Kang, S., Bader, A.G., and Vogt, P.K. (2005). Phosphatidylinositol 3-kinase
mutations identified in human cancer are oncogenic. Proc. Natl. Acad. Sci.
USA 102, 802–807.
Knight, Z.A., Gonzalez, B., Feldman, M.E., Zunder, E.R., Goldenberg, D.D.,
Williams, O., Loewith, R., Stokoe, D., Balla, A., Toth, B., et al. (2006). A phar-
macological map of the PI3-K family defines a role for p110alpha in insulin
signaling. Cell 125, 733–747.
Knight, Z.A., Feldman, M.E., Balla, A., Balla, T., and Shokat, K.M. (2007). A
membrane capture assay for lipid kinase activity. Nat. Protoc. 2, 2459–2466.
Kwon, C.H., Luikart, B.W., Powell, C.M., Zhou, J., Matheny, S.A., Zhang, W.,
Li, Y., Baker, S.J., and Parada, L.F. (2006). Pten regulates neuronal arboriza-
tion and social interaction in mice. Neuron 50, 377–388.
Lim, K.H., and Counter, C.M. (2005). Reduction in the requirement of
oncogenic Ras signaling to activation of PI3K/AKT pathway during tumor
maintenance. Cancer Cell 8, 381–392.
Liu, Y., Shah, K., Yang, F., Witucki, L., and Shokat, K.M. (1998). A molecular
gate which controls unnatural ATP analogue recognition by the tyrosine kinase
v-Src. Bioorg. Med. Chem. 6, 1219–1226.
Luo, J., Manning, B.D., and Cantley, L.C. (2003). Targeting the PI3K-Akt path-
way in human cancer: rationale and promise. Cancer Cell 4, 257–262.
Cancer Cell 14, 180–192, August 12, 2008 ª2008 Elsevier Inc. 191
Cancer Cell
PIK3CA Drug-Resistance Screen in S. cerevisiae
Mahon, F.X., Deininger, M.W., Schultheis, B., Chabrol, J., Reiffers, J., Gold-
man, J.M., and Melo, J.V. (2000). Selection and characterization of BCR-
ABL positive cell lines with differential sensitivity to the tyrosine kinase inhibitor
STI571: diverse mechanisms of resistance. Blood 96, 1070–1079.
Marone, R., Cmiljanovic, V., Giese, B., and Wymann, M.P. (2008). Targeting
phosphoinositide 3-kinase: moving towards therapy. Biochim. Biophys. Acta
1784, 159–185.
Odorizzi, G., Babst, M., and Emr, S.D. (2000). Phosphoinositide signaling and
the regulation of membrane trafficking in yeast. Trends Biochem. Sci. 25, 229–
235.
Paling, N.R., Wheadon, H., Bone, H.K., and Welham, M.J. (2004). Regulation of
embryonic stem cell self-renewal by phosphoinositide 3-kinase-dependent
signaling. J. Biol. Chem. 279, 48063–48070.
Quintas-Cardama, A., Kantarjian, H., and Cortes, J. (2007). Flying under the ra-
dar: the new wave of BCR-ABL inhibitors. Nat. Rev. Drug Discov. 6, 834–848.
Raynaud, F.I., Eccles, S., Clarke, P.A., Hayes, A., Nutley, B., Alix, S., Henley,
A., Di-Stefano, F., Ahmad, Z., Guillard, S., et al. (2007). Pharmacologic charac-
terization of a potent inhibitor of class I phosphatidylinositide 3-kinases.
Cancer Res. 67, 5840–5850.
Rodgers, E.E., and Theibert, A.B. (2002). Functions of PI 3-kinase in develop-
ment of the nervous system. Int. J. Dev. Neurosci. 20, 187–197.
Rodriguez-Escudero, I., Roelants, F.M., Thorner, J., Nombela, C., Molina, M.,
and Cid, V.J. (2005). Reconstitution of the mammalian PI3K/PTEN/Akt path-
way in yeast. Biochem. J. 390, 613–623.
Samuels, Y., Wang, Z., Bardelli, A., Silliman, N., Ptak, J., Szabo, S., Yan, H.,
Gazdar, A., Powell, S.M., Riggins, G.J., et al. (2004). High frequency of muta-
tions of the PIK3CA gene in human cancers. Science 304, 554.
Samuels, Y., Diaz, L.A., Jr., Schmidt-Kittler, O., Cummins, J.M., Delong, L.,
Cheong, I., Rago, C., Huso, D.L., Lengauer, C., Kinzler, K.W., et al. (2005).
Mutant PIK3CA promotes cell growth and invasion of human cancer cells.
Cancer Cell 7, 561–573.
Sarbassov, D.D., Guertin, D.A., Ali, S.M., and Sabatini, D.M. (2005). Phosphor-
ylation and regulation of Akt/PKB by the rictor-mTOR complex. Science 307,
1098–1101.
Shah, N.P., and Sawyers, C.L. (2003). Mechanisms of resistance to STI571 in
Philadelphia chromosome-associated leukemias. Oncogene 22, 7389–7395.
192 Cancer Cell 14, 180–192, August 12, 2008 ª2008 Elsevier Inc.
Shah, N.P., Nicoll, J.M., Nagar, B., Gorre, M.E., Paquette, R.L., Kuriyan, J., and
Sawyers, C.L. (2002). Multiple BCR-ABL kinase domain mutations confer poly-
clonal resistance to the tyrosine kinase inhibitor imatinib (STI571) in chronic
phase and blast crisis chronic myeloid leukemia. Cancer Cell 2, 117–125.
Shah, N.P., Tran, C., Lee, F.Y., Chen, P., Norris, D., and Sawyers, C.L. (2004).
Overriding imatinib resistance with a novel ABL kinase inhibitor. Science 305,
399–401.
Shaw, R.J., and Cantley, L.C. (2006). Ras, PI(3)K and mTOR signalling controls
tumour cell growth. Nature 441, 424–430.
Stauffer, F., Maira, S.M., Furet, P., and Garcia-Echeverria, C. (2008). Imi-
dazo[4,5-c]quinolines as inhibitors of the PI3K/PKB-pathway. Bioorg. Med.
Chem. Lett. 18, 1027–1030.
Stephens, L., Williams, R., and Hawkins, P. (2005). Phosphoinositide 3-kinases
as drug targets in cancer. Curr. Opin. Pharmacol. 5, 357–365.
Tugendreich, S., Perkins, E., Couto, J., Barthmaier, P., Sun, D., Tang, S., Tulac,
S., Nguyen, A., Yeh, E., Mays, A., et al. (2001). A streamlined process to phe-
notypically profile heterologous cDNAs in parallel using yeast cell-based
assays. Genome Res. 11, 1899–1912.
Vlahos, C.J., Matter, W.F., Hui, K.Y., and Brown, R.F. (1994). A specific inhib-
itor of phosphatidylinositol 3-kinase, 2-(4-morpholinyl)-8-phenyl-4H–1-benzo-
pyran-4-one (LY294002). J. Biol. Chem. 269, 5241–5248.
von Bubnoff, N., Veach, D.R., van der Kuip, H., Aulitzky, W.E., Sanger, J., Sei-
pel, P., Bornmann, W.G., Peschel, C., Clarkson, B., and Duyster, J. (2005). A
cell-based screen for resistance of Bcr-Abl-positive leukemia identifies the
mutation pattern for PD166326, an alternative Abl kinase inhibitor. Blood
105, 1652–1659.
Walker, E.H., Perisic, O., Ried, C., Stephens, L., and Williams, R.L. (1999).
Structural insights into phosphoinositide 3-kinase catalysis and signalling.
Nature 402, 313–320.
Ward, S.G., and Finan, P. (2003). Isoform-specific phosphoinositide 3-kinase
inhibitors as therapeutic agents. Curr. Opin. Pharmacol. 3, 426–434.
Zhao, J.J., Liu, Z., Wang, L., Shin, E., Loda, M.F., and Roberts, T.M. (2005). The
oncogenic properties of mutant p110alpha and p110beta phosphatidylinositol
3-kinases in human mammary epithelial cells. Proc. Natl. Acad. Sci. USA 102,
18443–18448.